Вы находитесь на странице: 1из 3

Transport-based Quantitative Photoacoustic Tomography

Lei Yao, Yao Sun, and Huabei Jiang


Department of Biomedical Engineering, University of Florida, Gainesville, FL 32611
Email: hjiang@bme.ufl.edu

Abstract: A method based on the radiative transfer equation (RTE) coupled with the
photoacoustic equation is presented. It provides quantitatively and significantly improved image
reconstruction for the cases where the photon diffusion approximation may fail.
©2010 Optical Society of America
OCIS codes: (110.5120) Photoacoustic imaging; (010.5620) Radiative transfer

1. Introduction
Photoacoustic tomography (PAT) is an emerging biomedical imaging technique for visualizing tissue structure and
function with excellent ultrasound resolution and excellent optical contrast [1-3]. The conventional PAT methods
provide only the distribution of absorbed light energy density. Recently several methods have been developed and
demonstrated that absolute absorption coefficient can be recovered when the conventional PAT is combined with the
photon diffusion equation (DE) [4-7]. These methods have certainly expanded the applicability of PAT. However,
the DE has certain limitations. For example, it fails to accurately describe light propagation in media where the
absorption coefficient is not small compared to the scattering coefficient, such as in haematoma, the cerebrospinal
fluid in the brain, the ventricles, or the subarachnoid space [8]. In addition, the variable anisotropy factor g also
affects the photon migration in tissues, which is not fully considered in the DE. It is generally believed that light
propagation in tissues is best modeled by the radiative transfer equation (RTE), which is formulated in phase space,
i.e., the space of positions and directions, making it computationally very expensive.
In this paper we describe a method for quantitative photoacoustic tomography (qPAT) based on the photon RTE
coupled with the Helmholtz photoacoustic wave equation. Considerable tissue-like phantom experiments are
conducted to evaluate the transport-based qPAT in comparison with the diffusion-based qPAT. The results obtained
show that the transport-based qPAT allows for clearly more accurate recovery of absolute absorption coefficient
images of heterogeneous media over the diffusion-based qPAT for the cases examined and provides considerably
improved image quality for cases where the photon diffusion approximation (DA) is invalid.
2. Algorithm
Our RTE based quantitative PAT reconstruction method includes two steps. The first is to obtain the map of
absorbed optical energy density through a model-based finite element reconstruction algorithm [9-11]. The core
procedure can be described as:
∇2 p( r , ω) + k 02 p ( r , ω) = ik 0 c0 βΦ( r ) C p , (1)
( ℑ ℑ + λI )∆χ = ℑ ( p
T T 0
− pc ) , (2)
where p is the pressure wave; k 0 = ω c0 is the wave-number described by the angular frequency, ω , and the
speed of acoustic wave in the medium, c0 ; β is the thermal expansion coefficient; C p is the specific heat; Φ
is the absorbed energy density that is product of absorption coefficient, µa , and optical fluence, Ψ (i.e.,
Φ = µa Ψ ); (
p 0 = p10 , p 20 , p M0 ) T
, ( ) T
pi0 and pic are observed and
p c = p1c , p2c , p Mc , and
computed complex acoustic field data for i =1,2, M boundary locations; ∆χ is the update vector for the
absorbed optical energy density, ℑ is the Jacobian matrix formed by ∂p ∂Φat the boundary measurement
sites; λ is the regularization parameter determined by combined Marquardt and Tikhonov regularization schemes,
and I is the identity matrix.
The second step is to recover the distribution of absorption coefficient from the absorbed energy density
obtained in the first step, and it is based on the iterative solution to the RTE:
     
(Ω⋅ ∇+ µ s + µa )ϕ(r , Ω) = µs ∫ n −1 ϕ(r , Ω′)Θ(Ω, Ω′)dΩ′ + q (r , Ω) ,
S
(3)
  
where µs is the scattering coefficient; ϕ(r , Ω) is the radiance; q(r , Ω) is the source term; Ω ∈S n −1

( )
denotes a unit vector in the direction of interest. The kernel, Θ Ω, Ω′ is the scattering phase function describing
 
the probability density that a photon with an initial direction Ω′ will have a direction Ωafter a scattering event.
In this study we assume that the scattering phase function depends only on the angle between the incoming and
 
( ) ( )
outgoing directions; thus Θ Ω, Ω′ = Θ Ω⋅ Ω′ . Here the commonly used Henyey-Greenstein scattering
v v
function is applied [12]: Θ ( Ω ⋅ Ω′ ) = (1 − g ) / 2π (1 + g − 2 g cos γ ) where γ is the angle between Ω′ and
2 2 

Ω, and −1 < g <1 .
 

( )
The optical fluence is related to the radiance by Ψ( r ) = ∫S n −1 ϕ r , ΩdΩ. If the incident laser source
strength and the absorbed energy density Φ are estimated in advance, µa distribution can be determined by the
following iterative solution procedure using finite element method [13,14]: (a) choose an initial value for µa , e.g.,
µa = 0.0001 mm −1 ; (b) compute the optical fluence Ψ using the RTE; (c) calculate the absorbed energy
density by Φ = µa Ψ ; (d) compute the error between the measured Φ (from the first step) and calculated Φc ,
c

and µa is updated using µa = Φ Ψ ; and (e) if the error is sufficiently small, then the iterative calculation stops;
otherwise repeat steps (b)-(d) until a small error is reached. µs and g are assumed as constant in this study.

3. Results and discussion


Two experiments were conducted with our PAT imaging system, described in detail elsewhere [11]. Briefly, an
Ti-Sapphaire laser generated a pulsed beam with a pulse repetition rate of 10 Hz and a pulse width <10ns which was
delivered to the top surface of a cylindrical solid phantom (Intralipid, India ink, distilled water and Agar power),
among which Intralipid served as the scattering medium and the India ink served as the absorption medium. A 1
MHz transducer was used to detect the acoustic signal at 120 positions along a circular path around the phantom.
Both the transducer and phantom were immersed in a water tank. In the experiments, we embedded a 3mm diameter
target in a 12mm diameter background. The absorption coefficient of the background was 0.01mm -1, and the
absorption coefficient of the target was 0.05mm-1 for both of the two cases. The reduced scattering coefficient was
0.05mm-1 throughout the media. The anisotropy factor g of the background and target is 0.59 and 0.88 for the two
cases, respectively.
Figs. 1a-d present the reconstructed absorption coefficient images for the two phantom experiments using the
RTE- (Fig. 1a for case 1 and Fig. 1c for case 2) and DE- (Fig. 1b for case 1 and Fig. 1d for case 2) based algorithms,
respectively. We see that the target can be clearly imaged by the two methods. The major difference between the
two methods lies in the ability of quantitatively resolving absorption coefficient. This difference in quantifying
absorption coefficient is clearly demonstrated in Fig. 2 which shows the recovered absorption property profiles
along a transact through the center of the target and background for the images shown in Fig. 1.

(a) (b) (c) (d)


Fig.1 Reconstructed absorption coefficient images

2
0.07 0.07

Exact Exact
0.06 PAT+RTE 0.06 PAT+RTE
PAT+DE PAT+DE

absorption coefficient (mm )

absorption coefficient (mm )


-1

-1
0.05 0.05

0.04 0.04

0.03 0.03

0.02 0.02

0.01 0.01

0.00 0.00
-4 -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4
Y position (mm) Y position (mm)

(a) (b)
Fig. 2 Recovered absorption coefficient profiles along (a) y=-0.5mm for case 1 and (b) y=0.0mm for case 2

We see that the DE-based method either underestimates or overestimates the absorption coefficient of the target
for a large µ a µ s′ ratio or increased g factor, while the RTE-based method always provides accurate
reconstruction of the absorption coefficient for both of the cases. We found that the average value of the recovered
absorption coefficient of target is 0.0489mm-1 and 0.0492mm-1 using the RTE-based method, which is in excellent
agreement with the actual value of 0.05mm-1, while the DE-based method gives a recovered target absorption
coefficient of 0.0565mm-1 and 0.0369mm-1 for the experimental cases. It is clear that the DE-based method presents
significant errors when the µ a µ s′ ratio is increased (case 1) or g factor is increased (case 2), while the RTE-based
method gives an error of below 3% for both cases.
In summary, we have presented a method for quantitative PAT based on the RTE. The experimental results
obtained suggest that the RTE based method offers significant improvement in reconstructing absolute absorption
coefficient of heterogeneous media for both the DE satisfied and unsatisfied cases with most significant
enhancement for the case involving a DE-unsatisfied background.

References
[1] G. Paltauf, J. Viator, et al., “Iterative reconstruction algorithm for optoacoustic imaging,” J. Acoust. Soc. Am. 112, 1536-1544 (2002).
[2] S. J. Norton and T. vo-Dinh, “Optoacoustic diffraction tomography: analysis of algorithms,” J. Opt. Soc. Am. A 20, 1859-1866 (2003).
[3] A. A. Oraevsky, A. A. Karabutov, et al., “Laser optoacoustic imaging of breast cancer in vivo,” Proc. SPIE 4256, 6-15 (2001).
[4] B. Cox, S. Arridge, K. Kostli and P. Beard, “Two-dimensional quantitative photoacoustic image reconstruction of absorption distributions in
scattering media by use of a simple iterative method,” Appl. Opt. 45, 1866-1875 (2006).
[5] Z. Yuan, H. Jiang, “Quantitative photoacoustic tomography: recovery of optical absorption coefficient map of heterogeneous medium,” Appl.
Phys. Lett. 88, 231101 (2006).
[6] L. Yin, Q. Wang, Q. Zhang, and H. Jiang, “Tomographic imaging of absolute optical absorption coefficient in turbid media using combined
photoacoustic and diffusing light measurements,” Opt. Lett. 32, 2556-2558 (2007).
[7] Z. Yuan, Q. Wang, and H. Jiang, “Reconstruction of optical absorption coefficient maps of heterogeneous media by photoacoustic
tomography coupled with diffusion equation based regularized Newton Method,” Opt. Exp. 15, 18076-18081 (2007).
[8] A. H. Hielscher, R. E. Alcouffe, and R. L. Barbour, “Comparison of finite-difference transport and diffusion calculations for photon migration
in homogeneous and heterogeneous tissues,” Phys. Med. Biol. 43, 1285-1302 (1998).
[9] Z. Yuan, C. Wu, H. Zhao, and H. Jiang, “Imaging of small nanoparticle-containing objects by finite-element-based photoacoustic
tomography,” Opt. Lett. 30, 3054-3056 (2005).
[10] H. Jiang, Z. Yuan, and X. Gu, “Spatially varying optical and acoustic property reconstruction using finite-element-based photoacoustic
tomography,” J. Opt. Soc. Am. A 23, 878-888 (2006).
[11] Z. Yuan, H. Zhao, C. Wu, Q. Zhang, and H. Jiang, “Finite-element-based photoacoustic tomography: phantom and chicken bone
experiments,” Appl. Opt. 45, 3177-3183 (2006).
[12] J. Heino, S. R. Arridge, J. Sikora, and E. Somersalo, “Anisotropic effects in highly scattering media,” Phys. Rev. E 68, 031908: 1-8 (2003).
[13] B. Cox, S. Arridge, K. Kostli, and P. Beard, “Two-dimensional quantitative photoacoustic image reconstruction of absorption distributions
in scattering media by use of a simple iterative method,” Appl. Opt., 45, 1866-1875 (2005).
[14] H. Jiang, Y. Xu, and N. Iftimia, “Experimental three-dimensional optical image reconstruction of heterogeneous turbid media from
continuous-wave data,” Opt. Express, 7, 204-209 (2000).